Tips for Developing And Administering Effective Organizational Surveys

Description

The help guide contains basic recommendations for developing, administering, analyzing, and reporting the results of organizational surveys. This information covers general issues and considerations relating to the survey process.

Surveys can provide valuable information if done correctly. They can suggest ways to increase productivity, boost morale, and reduce costs, while serving as vehicles for employees to communicate with management or participate in decision making. Before any of these goals can be accomplished, several tasks must be completed by the group or individual overseeing the survey process. These tasks include:

· Identifying the topics for the survey.

· Constructing the specific survey items and response scales.

· Identifying the participants of the survey using sampling techniques.

· Administering the survey

· Analyzing the survey data.

· Providing feedback of the results.

· Developing action plans.

Each of these tasks is explained in a separate section of this online help guide. Each section begins with questions that should be addressed and recommendations that summarize the issues.

Contents (click on a link to view that section)

Overview

Advantages and Disadvantages of Survey Research

The Survey Process

Focus on Issues

Identify Key Topics

Target Groups

Survey Design

Invitation to Participate

Question Development

Response Scale

Web Principles

Open-Ended Questions

Employee Characteristics

Pre-testing

Survey Administration

The Sample

Response Rates

Communication

Distribution

Confidentiality

Data Management

Data Cleaning

Sample Quality

Data Analysis

Probing Analysis of Open-Ended Questions

Survey Feedback

Action Plans

 

Overview

Organizational surveys are extremely useful tools for assessing the current state of any organization and can provide useful information for organizational change efforts, guiding strategic decision making, and enhancing productivity and employees' quality of work life.

Organizational surveys also communicate a message to employees. Ideally, surveys can signal to employees, customers or other stakeholders that their opinions are valued, that the organization is committed to them as individuals, and that their input is being used in decisions that affect them. Transmitting these messages can have a powerful influence on the desired outcome (e.g. performance, employee satisfaction, customer satisfaction)

When done incorrectly, surveys also can communicate the wrong messages as well. For example, surveys can show employees that their opinions don't matter if the organization doesn't use their input to make decisions or communicate the survey results.

Surveys can also be used to clarify the expectations the organization has of its members (awareness of organizational strategies, goals, values, missions) in the feedback process. For instance, surveys may be used to assess members' knowledge about their benefits and other aspects of compensation.

Advantages and Disadvantages of Survey Research

Using our system to conduct a survey is relatively inexpensive, although it may become costly if the project is large-scale and intended to serve a strategic purpose. If a survey is designed properly, the cost of the project is far outweighed by the usefulness of the information obtained from the survey. Alternatively, if the survey is not designed to meet an organizational purpose, it may be of little value. 

Surveys also have some potential disadvantages. Some people are skeptical that employees' feelings and values can be expressed as numbers. More importantly, surveys are specifically designed to be "standardized" Every employee receives the same questions and must respond using the same constraints. This often does not provide enough information to tell you why the employees answered the way they did. However, this information can be supplemented with information gathered from interviews with individuals or groups of employees or with open-ended questions present in the survey itself.

Ideally, a data gathering project in the organization uses a triangulation method in which managers and leaders are interviewed to get one perspective on a topic, a select number of focus groups are used to generate and validate items, and finally a survey is sent to a representative sample of employees.

The Survey Process

Before beginning the survey process, form a survey steering committee. The committee should be composed of representatives from different parts of the organization, but who have unique areas of expertise that relate specifically to the purpose of the survey. They will oversee the process and facilitate communication throughout the organization and report progress to the project sponsors.

The chief responsibility of this team is to develop a strategy for handling all phases of the survey project before data collection. This ensures that foreseeable events are addressed and increases the speed and efficiency of the latter phases of the project.

Effective surveys require communication at all levels of the organization. Employees need to see a link between survey results and actual organizational changes.

To accomplish its goal, the following steps should be followed in a survey research project.

Focus on the Issues

Design Survey

Administer Survey

Present Feedback Results to Management and Stakeholders

Develop Action Plans

Focus on Issues

Identify Key Topics

One of the critical decisions in constructing a survey is deciding what information the organization wants from the survey. This task requires a thorough consideration of the key topics of interest to the organization and the kind of information the organization wants on each topic. Identification of these key topics assists in the process of survey construction, analysis, and reporting. It also has implications for the choices of employee groups to complete the survey and the process of administering the survey.

Some possible uses of organizational surveys include:

· Pinpointing areas of concern: Surveys permit general evaluations of satisfying or dissatisfying aspects of work and can be used to identify possible areas for organizational development interventions.

· Observing long-term trends: Surveys can track changes in morale and other factors over time. This increases the organization's ability to respond to problems before they become crises.

· Monitoring a program's impact: Surveys can measure reactions, for example, to training programs.

· Providing input for future decisions: Surveys can provide the basis on which to make informed decisions on issues such as which benefits should be added to the current compensation program to meet employee needs, and which benefits are rarely used.

· Adding a communication channel: Surveys provide employees with an additional communication path to senior management.

· Performing organizational behavior research: Surveys guide strategies for changing employees' behavior.

· Assessing organizational change and development: Surveys can guide self-assessment and be used to assess the effectiveness of changes or to stimulate continuous improvement in the organization.

· Providing symbolic communication: Surveys send symbolic messages to employees such as, "We care:" Surveys can also send negative messages if the questions are loaded or seen as inappropriate, if feedback is not provided, if promised actions are not taken, if data are ignored, or if responses are covered up.

· Improving perceptions of fairness: Decisions that are made with employee input are more likely to be seen as fair than decisions that are simply "handed down" by management.

Once the general purpose of the survey is identified, the organization can begin to determine the specific topics to be covered. In some cases, it may be useful to conduct focus groups with a small group of employees to identify their concerns with specific topics. In others, it may be advisable to use standardized questionnaires.

It is important that the purpose of the survey be explicit, agreed upon by senior management, and communicated to employees before conducting the questionnaire. When no explicit purpose is communicated to employees other than, "We are interested in your opinion on several topics" employees often expect questions concerning their satisfaction with compensation, supervision, and work conditions. Then, if these questions are not included, employees may feel that the survey is of limited relevance and not put much effort in providing meaningful responses. They may not even return the survey.

If a specific purpose is explicitly communicated (e.g. to establish organizational values), employees are less likely to expect such general questions.

Target Groups

The target group is the group of individuals or constituencies that is affected by the survey. The information provided by the survey respondents must be applicable to the target group as a whole or the survey is of limited value.

For example, potential target groups within an organization include:

· The organization as a whole.

· Certain demographic groups.

· Certain job/functional groups.

· Work teams.

Surveying the entire organization has symbolic value. It communicates to employees that their opinions are important. Conversely, excluding certain groups of employees from a survey may send a negative or threatening message.

However, surveying the entire organization is often unnecessary for valid and reliable data. Surveying a sample group of employees allows the organization to conduct more surveys without "burning employees out" on responding to them.

In some cases, a certain demographic group will be of interest. For instance, an organization might be interested in how members of a minority group are faring in the organization.

Various job functions or work teams might be the focus of the survey. If one job function is of interest, then it may be advisable to limit the survey to that group and distribute the questions that are relevant to all employees to the entire workforce. It is generally not recommended to include questions that are applicable only to one group of employees on a survey of the entire work force.

Management Commitment

The survey steering committee may also want to include people whose support is needed to accomplish the purpose of the survey, even if their actual input is unnecessary.

Participation by top executives may increase their level of involvement in and commitment to the survey process. Including supervisors reduces the likelihood that they will perceive that the HR department or an external consultant is "going behind their back" and will help build their commitment to the project.

Top management commitment ensures that the importance of the survey is emphasized to managers and that first level managers are expected to follow through on changes deemed important by the survey.

The key to gaining top management commitment is effective communication of the benefits of the survey to the organization.

Group Size

Another decision the committee must make is how many individuals should participate. There are two principle advantages to larger numbers of respondents: more opportunity for organizational members to participate and reduced margins of error in the data.

Of course, the tradeoffs for increased participation are project cost and completion time, and potential burnout of the survey population.

If the company wants to compare the responses of two or more groups, then each group needs to be of similar size to have a similar margin for error.

Survey Design

The process of designing, implementing, and reporting the data back from an organizational survey is as important, if not more important, than the actual results. It is important to remember that the survey is a means to an end, not an end in itself.  Given this, there are several considerations to the construction of an effective survey:

· The invitation to participate.

· The instructions.

· The wording of the questions.

· The use of open-ended questions.

· The response format for the questions.

· Decisions about background questions.

Invitation to participate

Any survey should include a description of the purpose of the project. This description should establish a need for the project and attempt to overcome participant anxieties about the purpose of the project. Employees will want to know the answer to three questions:

· Why was I chosen to participate?

· What will be done with the data?

· How will the survey affect me?

These questions can be addressed in the invitation to participate and specific language addressing the 3 questions above should be included in the invitation whether by e-mail, regular mail, or by survey facilitators if the survey is administered in the workplace.

The invitation to participate should also include assurances of confidentiality. Employees should be assured that:

· The survey is not a test. (There are no wrong answers and honest answers are best).

· Only group level data are examined, not individual data. (For example, the data is analyzed in terms of how the entire Accounting Department responded, not in terms of any accountant's individual responses.)

· Individual employees cannot be identified from their responses to the questionnaire.

· Individual surveys are not made available to managers, or anyone else.

Another set of fears employees may have is that the survey will be used to take away resources from departments, cut staff, or cut compensation and benefits packages. These anxieties are likely to influence employees' willingness to complete questionnaires and may affect the honesty of their responses should they choose to participate.

If the organization uses the survey to "take away' something, employees' trust in senior management is likely to be severely damaged and the organization will have difficulty building trust in other organizational development efforts.

If possible, provide assurances that the project has no malevolent intent. This can be done through advance notification of the purpose of the project. Supervisors should also communicate the nature of the survey to their subordinates, and, if relevant, provide examples of how similar surveys have been used in the past.

If the survey is indeed intended to "take away" something (e.g., increased costs require some benefits to be discontinued), it is extremely important that the purpose of the survey be made clear from the beginning. Although the survey will certainly not endear management to the employees, soliciting employee opinions will ensure that the decision-making process and resulting decisions are more likely to be seen as fair.

Instructions

Surveys should include clear instructions for accessing and completing the survey on the Internet. Our system provides the opportunity to create customized instructions at the beginning of the survey and on each page of the survey. In order to combat the effects of fatigue induced by having to fill out lengthy surveys, it may be useful to break up the survey into sections. Each section should have an explanation of its purpose and instructions how to use the response scale.

The instructions can include tips like choosing the best answer (where applicable), where to write responses, and whether the questions refer to the employee or to employees in general.

The following is a set of sample instructions:

The following set of questions have to do with what you personally think about Company XYZ's work environment. For each item, indicate your agreement with the statement using the scale below.

1=Strongly Disagree; 2=Disagree; 3=Neither Agree nor Disagree; 4=Agree; 5=Strongly Agree

Include detailed instructions for the first section or two of the instrument (clarifying all the steps in the task such as, "...then select the appropriate response option using the radio button/drop down menus provided……..") and then use brief instructions for subsequent sections.

Survey designers should be careful to avoid leading instructions. Leading instructions refer to instructions that structure the respondent's perceptions in such a way that one response or another is preferred.

As an example, compare these two sets of instructions:

A recent survey showed that most bridges in America are in serious need for repair. If these problems are not addressed, the lives of American motorists may be in danger. Do you support the federal government spending additional money to repair bridges?

vs.

It is important to maintain American bridges to ensure the safety of motorists. However, to accomplish this purpose would require a 1 % income tax increase at a time when many Americans feel that too much of their money is taxed. Do you support the federal government spending additional money to repair bridges?

One might expect that for some people, particularly those without strong opinions, the wording of the first set of instructions may influence their answer to the question.

Question Development

A good guideline for organizational research is that "you only get what you measure" and, conversely, "if you don't want to know, don't ask."

The data obtained from a survey are only as good, specific, or as broad as the questions in the survey.

Here are some guidelines to keep in mind when constructing questions:

· Keep questions short. Unless a detailed explanation or clarification of an issue is absolutely necessary keep questions short.

· Avoid ambiguity. Questions should be specific, avoiding generality. For example: "On a scale from 1 to 5, how popular is President Bush at this time?" This example begs the question, "popular with whom?" Some respondents will answer an item like this in terms of how they, personally, would rate him; others will answer in terms of popularity with their peers, and still others will answer according to their recollection of the last opinion poll.

· Avoid unfamiliar terms and jargon. For example, "Does your department support expatriate research?" Terms such as expatriate may not be well understood by typical employees. When a term that is not in common usage must be used, it is wise to precede the item with a brief explanation of the unfamiliar term.

· Avoid requiring inaccessible information. An item may use terms that are familiar to employees while requiring information that they would not know or do not have access to. For example, "My company offers benefits that are superior to those offered by other companies" assumes that employees are familiar with both the benefits their employer offers and those offered by other companies.

· Avoid hypothetical questions. Questions that begin with "What would you do if…" are difficult for employees to answer. They require greater mental effort on the part of the employee, and take longer to answer.

· Be cautious with questions that ask employees to rank order lists. Ranking can be a challenging task. Generally, employees should not be asked to rank order more than four or five items. Beyond that point, employees may begin giving arbitrary responses just to get past the item.

· Write questions in clear and common language. People's ability to recall the past is limited. The more current and specific the question, the better. If a recall question is necessary, (e.g., "Over the past six months, I haven felt harassed at work") the time frame should be as recent as possible, and not over six months unless the reference is to a major event (e.g., marriage, buying a car, changing jobs).

· Write questions in clear and common language. Use words that all employees understand. Most word processing programs include reading level indices that indicate the level of difficulty of the wording of the questions.

· Specific questions are likely to be more useful than general questions. Generally, more specific questions are more useful than broad or vague questions. Employees also find specific questions easier to answer.

For instance, it is more informative to ask employees how satisfied they are with specific elements of their job than it is to ask employees about the job in general - particularly if the purpose of the survey is to identify specific components of the job that might be changed. If the purpose of a job satisfaction question is to simply have a general assessment of satisfaction, the more global question may be sufficient.

· Do not use questions that could cause anxiety. Some questions may cause anxieties about employees' futures. Where possible, avoid these questions (e.g., "I would feel nervous if my organization was laying off employees.")

· Offer a no opinion option. If employees are forced to offer an opinion, they probably will. Although offering a "don't know" or "no opinion" option for attitudinal or perceptual questions is sometimes perceived as losing information, forcing people to express an opinion may lead to misinterpretations of the data. Forcing a choice distorts the responses of employees who genuinely have no opinion, are genuinely unsure, or who genuinely do not care.

One solution is to include a second question that allows employees to express how important the issue is to them. The combination of agreement and importance can provide a great deal of meaningful information about employees' perceptions.

· Use positively worded questions. Although not all researchers agree, we recommend using positively worded questions for organizational surveys measuring workplace perceptions.

The issue is whether the response to a positively worded question such as, "I am satisfied with my working schedule" produces comparable information to a negatively worded question on the same topic such as, "I am not satisfied with my work schedule." Theoretically, strong agreement with the first question should mean the same thing as strong disagreement to the second.

There are at least two reasons why this may not be the case. First, some respondents may agree with any statement that is presented to them. If employees respond in this way, the results of the survey are difficult to interpret. The second reason concerns carelessness. If employees miss the "not" in the second question, they may misinterpret the question.

Stress the importance of reading each question carefully in written (and verbal) instructions.

· Do not use "double-barreled" questions. Double-barreled questions are questions that tap more than one topic. These questions can introduce ambiguities into responses. An example of a double-barreled question would be, "I am happy with my pay and benefits" If employees disagree with this statement, they may be expressing dissatisfaction with their pay, their benefits, or both. Generally, the use of conjunctions like "and" or "or" should be carefully considered. Double-barreled questions should be separated into two questions.

· Avoid questions that everyone will agree or disagree with. The purpose of the questionnaire is to provide information about the target group of individuals. To be useful, a question must be able to distinguish among groups of respondents. For example, a performance appraisal system where every employee is rated as outstanding will not help the organization determine which employee should be promoted. Questions should be written to measure a topic and distinguish among employees.

· Use multiple measures of the same topic. If a survey is designed to assess attitudes or opinions about issues that are subject to some interpretation by employees, then the survey should include multiple questions on each topic. This ensures that the responses are a function of the employee's perceptions of the issue and not a function of the idiosyncrasies of a single question. When multiple questions are asked, the average of the answers to these questions produces a "total score" for that topic. It is this total score that should be used to evaluate the issue. The proper approach to this problem is to ask multiple related questions, while avoiding asking the same question over and over. For example, if the topic of interest is fairness of compensation, some sample questions might include: "I am paid what I am worth", "The pay offered by my organization is similar to that offered by other organizations", and "I make the average salary or above for my pay grade."

· Keep the survey to a reasonable length. Generally, the survey should take no more than 30 to 45 minutes of an employee's time to complete. If it takes longer, the employee may become bored and begin responding in a careless fashion.

Response Scale

The choice of a response scale is a function of a number of statistical and practical concerns. The most important point is to choose a response scale that matches the information requested in the question.

Two general rules are described below:

· Use an odd number of response options (3, 5, 7). This is particularly relevant where a middle value can be used to assess "no opinion or neither agree nor disagree"

· Use no more than seven points in the scale. Some research suggests that in most cases people cannot distinguish more than five to seven points. Little increase is gained in precision beyond seven points.

Principles for Constructing Web-Based Surveys

Principle 1. Introduce the web questionnaire with a welcome screen that is motivational, emphasizes the ease of responding, and instructs respondents on the action needed for proceeding to the next page.

Principle 2. Begin the web questionnaire with a question that is fully visible on the first screen of the questionnaire, and will be easily comprehended and answered by all respondents.

Principle 3. Present each question in a conventional format similar to that normally used on paper questionnaires.

Principle 4. Limit line length to decrease the likelihood of a long line of prose being allowed to extend across the screen of the respondent's browser.

Principle 5. Provide specific instructions on how to take each necessary computer action for responding to the questionnaire.

E.g., employees may not know:

· Radio buttons require clicking an alternative button in order to erase a previous answer, while check boxes require a single-click in order to erase them.

· Respondents may not know how to operate a scroll bar in order to see the entire question or the next questions.

· It may not be clear how use of the clicker (or mouse) differs from use of the return key.

· It may not be clear to some respondents what a drop-down menu is, and how to access the hidden categories.

· It may not be clear how open-ended answers are to be entered, or that in some cases there is far more space than shows on the screen.

Principle 6. Provide computer operation instructions as part of each question where the action is to be taken, not in a separate section prior to the beginning of the questionnaire.

Principle 7. Do not require respondents to provide an answer to each question before being allowed to answer any subsequent ones.

Principle 8. Construct web questionnaires so that they scroll from question to question unless order effects are a major concern, large numbers of questions must be skipped, and/or a mixed-mode survey is being done for which telephone interview and web results will be combined.

Principle 9. When the number of answer choices exceeds the number that can be displayed on one screen, consider double-banking with appropriate navigational instructions being added.

Principle 10. Use graphical symbols or words that convey a sense of where the respondent is in the completion progress, but avoid ones that require advanced programming.

Principle 11. Be cautious about using question structures that have known measurement problems on paper questionnaires, (e.g., check-all-that-apply and open-ended questions).

Open-Ended Questions

It is advisable to include at least a few open-ended questions in any survey. At the minimum, a "comments" section will provide respondents the opportunity to voice their opinions on issues that they feel are important but that were not addressed in the survey.

Specific open-ended questions such as, "What advice would you give your CEO to improve the performance of the company?" are more effective at eliciting specific and useful answers than are vague or general questions. Open ended questions are especially effective in Web surveys; research has demonstrated that survey respondents tend to provide longer, more complete answers to open-ended questions in web surveys as opposed to the paper-and-pencil format.

Open ended questions are also often used as compliments as opposed to alternatives to structured survey items. As a follow-up to important structured items, an open-ended question can point out unanticipated interpretations of questions and variations in the actual meanings in responses. For example, a rank-ordering question about job satisfaction may show "income" to be the most important factor in satisfaction. A follow-up open-ended question may reveal that some employees meant "high income", while others meant "steady income".

The principle drawback to open-ended questions concerns the increased difficulties associated with incorporating the comments into the quantitative data analysis. This issue is addressed in the Data Analysis section.

Employee Characteristics

In many cases, it is useful to include questions on employee background characteristics. These questions concern personal employee characteristics (e.g. age, gender, marital status) or job/work characteristics such as geographical location, functional unit (e.g. sales, marketing, finance, production), supervisory unit, and hierarchical level.

These questions are useful to break down the respondents into sub-groups of interest For instance, on an attitudinal survey, the organization might be interested in whether reactions differ by factors such as gender, shift, location, exempt vs. non exempt, full-time vs. part time, and job classification. These issues need to be identified in the planning stage because, if they are not, the organization will usually be unable to identify the characteristics of a given respondent during the data analysis.

However, the survey should only measure employee characteristics if they are important for a particular survey purpose. Background measures are often seen as intrusive or unnecessary by employees and may be viewed as a way to identify them (particularly if many different characteristics are identified). For example, the responses of the only female supervisor who works on the second shift can be easily identified if the survey includes questions concerning gender and shift. In a setting where some employees are unionized and others are not, including a question that indicated union status prior to collective bargaining might be interpreted as an unfair labor practice.

Employee characteristics provide more useful information to the organization when there are a large number of people in each category. For instance, if 98% of the respondents are male, then a question on gender provides little information about the responses of most employees. If 45% of the target group are part-time and 55% are full-time, this may be an informative comparison question.

Legal Issues

Generally speaking, it is best to avoid asking any question that deals with race, color, physical appearance, disability, citizenship/national origin, or religion. Such questions can be seen by employees as offensive and invasive of their privacy.

Survey questions should be constructed on a need-to-know basis. That is, any question that is asked should have a specific, defensible purpose. Survey questions that touch on any of the above topics should be examined by an attorney familiar with labor law at local, state, and federal levels.

Pre-Testing

Organizational survey efforts often produce less than optimal information because of ambiguities in the instructions or questions. One way to overcome some of these problems is to pre-test the survey.

Pre-testing involves conducting a "dry run" with a small group of the target sample. They complete the survey while talking through the process with survey facilitators. This procedure involves no more than an hour or two of additional work and can help avoid numerous problems and embarrassments later in the process, as well as additional hours of work

Pre-testing is most effective when the group actually completes the survey rather than simply looking it over. Typically, the group will ask for clarification on the instructions ("Should I respond with how I feel or how people in general feel?), point out typographic errors, or note the need for additional questions on specific topics related to the purpose of the survey ("People on my team complain about vacation policies, why isn't that included on the list of benefits?").

A typical pre-testing session also includes a group discussion of the survey after the questions have been completed by all group members. Pre-testing may also help the organization avoid offensive questions or questions of a particularly sensitive nature.

The organization should use an initial group to identify major problems and then use a second group to double-check the corrections. These groups should represent the intended target group. Thus, if the target group is non-exempt retail workers with a high school education, it would be inappropriate to pre-test the survey using senior level managers.

Survey Administration

The Sample

Organizations must compete with tele-marketers, political opinion surveyors and social and psychological researchers. So, the organizational survey must effectively compete with the myriad of others clamoring for employees' time and attention. To accomplish this, the survey must be perceived as relevant to the employee and the survey task must be seen as relatively easy and non-intrusive. These needs highlight the importance of careful question construction and provision of feedback.

The actual participants in the survey are sometimes called the survey sample. One factor influencing the quality of the information obtained from a survey is the extent to which the sample of participants is similar to the entire group of individuals that is the intended target of the survey.

For instance, a benefits survey that only sampled single and unmarried individuals with no children would obtain very different information about individual preferences for various benefits programs than a survey that included married individuals and/or individuals with children. To maximize the quality of the information obtained from the survey, the organization must obtain a high quality sample.

Generally speaking, there are four sample construction strategies:

· Census

· Random sample

· Stratified random sample, and

· Mostly random sample

In a census, all relevant target members of the organization (or all members of a given business unit or job class) complete the survey. This method ensures that all members of the organization have the opportunity to express their opinions and is therefore, preferred.

Since an organizational census may be prohibitively costly, one alternative is to obtain a random sample from the target groups of interest. Random sampling simply refers to choosing a subset of a larger group where every member of the larger group has an equal chance of being chosen for the sample (i.e., participation is determined by chance).

Organizations often use stratified random samples to ensure that the sample has similar characteristics to the entire organization.

In this method, individuals are chosen at random from within each subgroup of interest. The proportion of the sample with a given characteristic should be as close as possible to the proportion in the organization. This ensures that the characteristics of the sample are representative of the organization as a whole and, therefore, ensures that conclusions drawn from the data apply to the organization as a whole. For instance, an organization with 75% non-exempt and 25% exempt employees might choose to obtain a random sample from within each of these groups. So, 75% of the sample would be non-exempt and 25% would be exempt. The stratified random sample strategy is more likely to produce a representative sample than the simple random strategy

The final method is to use a mostly random sample where individuals are chosen at random but all individuals with a particular characteristic are chosen.

This technique is useful where the responses of a very small (or very important) subgroup are of particular interest. For instance, in an organizational diversity survey, the sample might consist of randomly chosen members of the organization and all minorities (to ensure that their views are represented). Alternatively, the organization might choose to include all first line supervisors and a random sample of their subordinates. 

Each strategy produces samples with different respondent characteristics. These differences may influence interpretation of the data.

The census method is preferred, unless administering surveys to all members of an organization or a department is too costly.

A random sample of the entire business unit should provide a reasonably representative sample of the organization unless there is some reason to expect that the response rate will differ for one group of individuals.

Stratified random samples increase the likelihood that the sample will be representative, but involve more careful planning than the simple random sample method.

Finally, the mostly random sampling method ensures that all members of a particular group have the opportunity to present their views. In practice, organizations may choose to combine these methods.

Response Rates

From the perspective of the organization, the individuals who respond to the survey must be reasonably similar to the target group of interest. If the individuals completing the survey are not reasonably similar to the target group, conclusions drawn from the survey may not be appropriate for the target group.

If an organizational census is taken, employees' trust in the procedure is critical, as that influences the honesty and thoughtfulness of their responses. If a sampling strategy other than the organizational census is adopted, the critical issue is the survey response rate. The response rate refers to the proportion of people who actually completed the survey of those who were given the opportunity.

Response Rate = Number of completed surveys/Total surveys distributed

The smaller the response rate, the less likely the results are to be applicable to the target group. That is, the lower the response rate, the less likely the characteristics of the respondents are to be similar to those of the target groups. So, maximizing the response rate is a critical task for the survey steering committee.

Communication

To encourage participation, let employees know about an upcoming survey. At a minimum, the communication should include:

  • Advance e-mail notifications about the upcoming survey
  • Postings on the company bulletin board
  • Articles in a company newsletter
  • Memorandums to all employees

It is also be beneficial to use:

  • Telephone messages as reminders to encourage participation.
  • Discussions with first level supervisors and employees before data collection.

Any form of advance notification that increases employees' understanding of and interest in the survey is likely to increase the response rate.

Distribution

Onsite Surveys

One of the most effective survey distribution method is the onsite group survey. Generally, groups of employees meet and complete the survey while they are at work.

The advantages to the onsite group survey include its symbolic value and the increased likelihood that employees will participate. If the survey is administered during regular working hours, it signals to employees that the survey is important. Further, increasing participation increases the quality (representativeness) of the sample.

The principle disadvantage of the onsite method is the compensation costs accrued because employees typically complete the survey during regular working hours. Some companies may encounter logistical difficulties with removing groups of employees from their jobs for a brief period of time.

Obviously, onsite surveys must be carefully planned, especially in cases where the jobs of interest are highly dependent on each other or when individuals have production deadlines to meet.

Facilitators should be present during administration to explain the purpose of the survey and address questions employees might have during completion of the survey.

Internet Surveys

The Internet is fast becoming the medium of choice for conducting surveys. The primary reason is that Internet surveys are the most cost and time effective of all the administration methods. Internet surveys eliminate the need for printing, postage, and data entry costs and can significantly reduce the time needed to administer the survey. In addition, Internet surveys reduce the likelihood for human error to taint the results of the survey by eliminating the need for manual data entry. Below are a list of the advantages of Internet-based survey administration:

  • Improved Speed: Internet-based surveys eliminate the time needed for manual processing, printing, mailing, waiting for surveys to be returned, and time necessary for manual data entry and data analysis. Speeding up the assessment process enables a faster response to stakeholder needs.
  • Improved Reliability: Inevitably, when entering data manually and analyzing data, there is increased opportunity for human error. Internet surveys eliminate the possibility for human error by automating the data entry and analysis functions.
  • Improved Manageability: Internet-based survey management systems, such as our system streamline the assessment process and provide a virtually hands-free process.
  • Improved Security: Internet-based surveys remove threats to confidentiality by eliminating the need for paper surveys to be passed through the organization. Furthermore, with us hosting your survey products and data on our secured servers, it will be virtually impossible for any unauthorized users to access the system.
  • Decreased Costs: Over the long-term, organizations will achieve significant cost savings by eliminating printing, mailing, data entry, and data analysis costs from future survey initiatives.

Mail Surveys

The most traditional alternative to the onsite group survey is the mail survey. Mail surveys are easier to administer than on-site surveys because employees can complete the survey on their own time.

Mail surveys have three primary disadvantages that must be overcome for the survey to be effective:

There are logistical difficulties associated with both distribution and return of the surveys. The distribution process must be planned carefully and includes steps such as arranging a return address, arranging return postage, stuffing envelopes, and ensuring that employee mailing lists are up to date.

Employees may perceive the survey to be unimportant to the organization if they are asked to complete the survey on their own time.

The organization runs the risk of obtaining a low response rate potentially creating problems with the interpretation of the findings.

If the organization chooses to conduct a mail survey, each of these difficulties must be addressed.

There are several methods to increase the response rates to surveys:

· Pre-notification. Notifying participants prior to survey administration via e-mail, telephone, or regular mail so they have a "heads up" about the survey project.

· Reminders. The most common reminders are an e-mail or postcard reminding the employee to turn in the survey and providing them with instructions on how to access the survey.

· Incentives for completion. These can be included with the survey (e.g., thank you gift certificates that can be printed) or may be provided as a condition of completing a survey. Some companies use raffles offering a prize to employees who complete the survey. The raffle is a useful method, however, it can have unintended consequences. Ideally, the raffle entry form should be completed separately from the survey, otherwise, employee confidentiality concerns are raised as it is easy to identify people who include their name or identifying information with the survey!

Finally the issue of equal access should be taken into consideration. For instance, employees who do not (or cannot) check their e-mail on a regular basis, may miss the opportunity to participate in an email survey, and arrangements should be made to give them an opportunity to complete the survey. One solution would be to make a PC or Kiosk available in a open space (e.g. lunch-room) within the organization where employees may access the survey.

Confidentiality

One of the principle strengths of the organizational survey is the ability for employees to provide confidential assessments of the organization. When employees believe that their responses are confidential, they are more likely to provide accurate information.

If any administrative procedure is seen as untrustworthy by employees, their confidence in the privacy of their responses is likely to be undermined. One important principle is that supervisors should never handle or see the individual responses of their subordinates during any phase of the project. This is a critical concern and, if not followed, can seriously undermine the effectiveness of the survey

Conversely, organizations that make a strong effort to develop and maintain trust with their employees are likely to obtain useful data from the survey In any case, it is important to remember that the administration of the organizational survey is an act of symbolic communication.

Data Management

Once completed surveys are obtained, they must be entered into a data management system in order to process the data.

The data management procedures should be decided upon long before the survey is distributed. Certain data management procedures influence the design of the questionnaire itself as this may influence decisions about the personnel needed to conduct the analysis.

Statistical issues should be addressed by those individuals responsible for conducting the data analysis. Hiring an experienced data analyst to lead this phase of the project is recommended. If internal staff are used, they should be consulted before data collection to ensure that the organization has access to the software and hardware necessary to conduct any analysis of interest.

Data Cleaning (Mail surveys)

The first phase of any data analysis is data cleaning. Data cleaning simply means confirming that the data used in data analysis are sensible and match employees' actual responses.

The first phase of data cleaning takes place during data entry The individuals who are responsible for entering the data into the data management program should be trained to look for "out-of-range" values or (such as a "6" on a 1 to 5 response scale) or patterns of responding that suggest that the respondent did not actually attempt to answer the question. For instance, if the respondent assigned the value "1" to every question, the validity of that person's responses may be questioned (particularly if the answers are logically inconsistent).

The data cleaners should also watch for patterns of responding that have little to do with the questions. For instance, some individuals will simply fill out a pattern of numbers (such as 123212321232123) and not bother to read and respond to the questions.

The following is a list of common errors made by respondents:

· Multiple marked items. When the respondent provides two or more responses to each item.

· Cross-marked items. When several responses are circled or checked

· Changes to survey items. When the participant rewords the item and then answers that item rather than the intended item.

· Middle marks. When the respondent creates his/her own response options.

· Misaligned responses. When the respondent gives the "right" answer to the "wrong" question. This is likely to happen when optical scanned response sheets are used and the respondent inadvertently skips a space on the answer sheet (e.g., answering the first question in the second space).

Two additional steps are recommended once the data have been entered. The first is to examine frequency distributions (the percentage of people who gave each response) for each individual item for out of range values.

For instance, if the response "8" shows up in the frequency distribution on a seven-point response scale, that entry is probably incorrect. These errors should be identified and corrected before data analysis. (This highlights the importance of holding on to the original surveys.) This step is a minimum check and should be conducted in any circumstance. However, this step cannot identify incorrectly entered responses that are not out-of- range values.

The second step, that is more laborious, is to confirm the data entry. This involves physically double-checking the correctness of the values entered into the program. In a small data set this is easily performed for all surveys. In a larger data set, the data managers may decide to randomly check a small number of surveys to assess the likely error rate. If a random check of 5 to 100/0 of the surveys uncovers many errors, the data should be reentered.

For most questions, a "missing data code" should be developed. This code indicates that the respondent did not provide an answer to a question. This might be because the respondent committed an error, because the response was illegible, or simply because the question was left blank

One convention with five-point or seven-point response scales is to use the value "9" to denote missing data or a "99" to denote missing data on two digit items (e.g., age). Alternatively, several missing data codes could be developed to denote different forms of missing data. The common sense point is not to develop codes for the missing data that could also be meaningful responses.

Finally, some organizations use bubble sheets to eliminate the need for staff and reduce the possibility of human error. Bubble sheets allow the organization to electronically scan the data rather than to use manual data entry. If the organization has the technology to use this method, it can result in time and cost savings. However, it is important to remember that mistakes can be made in this process and that data should still be checked for accuracy. Whenever possible, computer technology should be used to eliminate manual data entry.

By using our system to develop and administer your organizational surveys, you will be able to avoid the potential for data entry errors. It is still necessary to examine frequency distributions for each survey item, and examine the data file for response patterns that are suspect.

Sample Quality

The second phase of data analysis involves assessing sample quality. This is particularly important when a sampling strategy other than the organizational census is used, but is useful any time that less that 100% participation is obtained.

Assessments of sample quality involve comparing the characteristics of the sample to known values of the target group from the organization's records. For instance, if the participants' were asked to provide their gender on the survey, the proportion of males and females in the sample can be compared to the proportion in the target group. If the values are substantially different, it may suggest problems with the quality of the data. These differences are of particular interest when individuals' responses to other questions are expected to differ on the comparison variable.

As an example, consider an organization where 50% of the workers were part-time and the part-time workers had a different (or no) benefits program than the full-time workers. A survey of benefit satisfaction that used a sample of 90% full-time workers would be likely to produce data that led to inappropriate conclusions about the employees as a whole.

The ability to assess sample quality is one advantage to including background characteristics on the survey. However, it is important to keep in mind that particular background questions may raise employees' concerns about their ability to be identified or raise questions about the "real" purpose of the survey. Sample quality concerns are alleviated when all members of a particular group of individuals are sampled.

Data Analysis

The data analysis phase of the survey may vary substantially depending on the purpose of the survey project. If the purpose of the survey is assessment (of benefits preferences, attitudes, etc.), the analysis will be fairly straight forward. If the purpose is to test a theoretical model that might be used to guide organizational change, the analysis may be quite sophisticated.

One general principle is to begin with a consideration of the strategic purpose of the survey. This provides the user with some sense of the information that will be of interest to consumers of the data and provides a sense of what types of analysis should be conducted.

For most organizational purposes, a general guideline is to keep the data analysis reasonably simple. Having "too many numbers" can often overwhelm people and cause the project stray from the goal of communicating knowledge.

Generally all data analysis will begin with frequency analysis. This involves constructing frequency distributions that describe the percentage of people who gave each possible response to a given question.

Our system provides various options for presenting your survey data. The final presentation format may vary according to organizational purpose. Depicting the data using a bar chart or pie chart may increase the visual appeal of the data and may make the results easier to "sell" to the organization. However, the goal is to communicate information. An overwhelming number of visually appealing charts may not be as effective as a small number of numerical tables that address the key points of the analysis.

A second way of presenting the same data would involve calculating the mean (average) value for each question. That is, if the same two questions above were scored on a five-point response scale (where strongly agree = 5) and the distribution of responses was the same, then if the sample consisted of 1,000 employees, the average score for question #1 would be 3.45 and the average response for question #2 would be 2.50. These averages provide us with one number that captures the entire sample on each question. In this case, it provides us with evidence that employees tend to be satisfied with their pay but many have poor relationships with their supervisors.

Once again, the user must keep in mind the of the analysis when choosing which format will be most effective.

Often, in statistical analysis, we are interested in the margin of error of a statistical estimate. That is, if our sample consists of less than 100% of the population of interest (e.g., the whole organization) we want to say, with some degree of confidence that the true value should lie within a certain range. This range is often called the confidence interval.

The confidence interval is calculated as [the mean +/- two standard errors]. (The standard error for an estimate is readily obtainable by most statistics software packages.) So, if the average score for question #1 was 3.45 and the standard error was .30 we could use the confidence interval to say that we are 95% confident that the "true" value is somewhere between 3.15 and 3.75.

Because of the way the standard error is calculated, using larger numbers of employees leads to smaller standard errors. So larger samples provide us with more precise estimates.

Probing

After seeing the data in the previous example, the organization might be interested in probing the sources of pay satisfaction a little further.

One simple technique is to conduct a frequency breakdown by filtering the data. This analysis simply involves conducting the frequency analysis for subgroups of interest.

For instance, the organization might suspect that part-time workers are less likely to feel satisfied with their pay. In this case the same frequency analysis could be conducted separately for full and part- time workers.  This type of probing can be conducted using the filtering capabilities within Our system, or by exporting the data file for further exploration in another program.  

Analysis of Open-Ended Questions

The chief drawback to open-ended questions is the difficulty of summarizing the responses in numerical form. Because of these difficulties, the written responses to open-ended questions often may not be systematically analyzed and therefore, a potentially valuable source of information is lost.

While analyzing written responses is challenging, some procedures have been developed. The most common approach to summarizing written data is generally termed "content analysis:'

In its most basic form content analysis consists of two steps. First, each written response is typed on a separate page. Then, one or more individuals record the "themes" in each response. The frequency with which each theme occurs in the written responses can provide some indication of the importance of the issue to members of the organization (e.g., 15% of the respondents indicated that alternative work schedules should be considered). These themes must be specific enough to communicate the nature of the comment and general enough to apply to many different individuals' responses.

For instance, mentions of the performance appraisal system could be either positive or negative. In this case, the "direction" of the comments has important implications for interpretation of the data and should be incorporated into the coding scheme.

Two issues should be kept in mind when conducting content analysis. First, using only one individual to identify the themes runs the risk of letting that individual's biases influence the interpretations of the data. To overcome this problem, two or more individuals can examine each comment and the extent to which they agree on the coding can be examined. This procedure adds additional time and costs to the data analysis but can increase the quality of the data provided.

The second issue is that individuals with strong opinions are more likely to take the time to express them in writing. Further, other individuals with strong opinions may not express them on the survey. These constraints should be kept in mind when reporting the frequency with which themes are mentioned.

For example, it is useful to note the percentage of people addressing a particular theme as a function of the number of people participating in the survey and as a function of the number of people offering opinions. As an example:

Of the 1,000 individuals who returned surveys, 250 (25%) offered written suggestions on how the company could improve performance. Of these, 125 (50%) indicated that additional customer service training might be useful while 50 (20%) expressed concerns about the performance assessment system..."

Survey Feedback

A survey that is constructed and administered with the utmost attention to quality is meaningless if the results are not effectively communicated to managers and survey participants. Thus, the process of providing feedback to employees is critical. Two critical decisions are, "Who should see the results?" and "How will the results will be communicated?"

Generally, all employees who participate in the survey should receive some form of feedback. Being open and honest with employees builds trust, facilitates the identification of potential solutions to problems, and generates commitment to the change process.

The specific format of the feedback session may vary according to organizational circumstances and resources (e.g., depending upon the time, and human and material resources the organization can devote to the feedback sessions). The purpose of the survey will provide useful information for structuring the feedback session.

In the top down, or waterfall method, results are first communicated to top managers who then pass results down through progressively lower levels of the organization. This technique is useful for results that have implications for the entire organization.

Bottom up or bubble up feedback may be useful when local issues are the focus (e.g., quality of work life within departments or teams). The nature of the data presented to employees may differ in form depending on the communication media (e.g., written reports, lectures, discussions).

A general guideline is that the survey researcher should ensure that the findings of interest to the employees are presented in a way that they can understand, without heavy reliance on statistical terminology or pages of numbers with no clear descriptions of the implications of the data.

The more actively involved employees are in the feedback process, the more effective the survey process will be.  On the low end of activity, survey feedback may consist of providing employees with a written report summarizing the data.  Moderately engaging feedback procedures include survey workshops or focus groups of employees where the results are described to employees who then have the opportunity to discuss the results and ask questions.

On the high end, and the most effective, is a feedback process where executives meet with employees to present the results, discuss the implications of the results, explore the underlying reasons for the results, and generate solutions to problems. This process maximizes employee involvement thus increasing the information value of the survey and provides an effective means of building trust and commitment and opening communication channels.

Managers can maximize involvement by conducting meetings with individual employees. This is an effective technique in smaller organizations or for smaller departments. However, group meetings may encourage some employees to speak out and, if properly conducted, these sessions allow group "brainstorming" to identify underlying causes and solutions to problems.

Trained facilitators are critical to the effectiveness of feedback. First line managers may be trained to conduct the sessions (rather than consultants). Using first line managers encourages "ownership" of the data. However, if the results of the survey are overwhelmingly negative, employees may be unwilling to discuss the results openly with their supervisor (particularly if the supervisor is perceived to be the problem). Further, asking the supervisor to contend with heavy employee criticism may be asking too much. Given no extreme problems in the data, using trained managers is generally recommended.

Feedback sessions are most effective when the aim of the feedback meeting is to identify the underlying causes of the patterns of responses. The group should have the task of focusing on identifying the underlying causes of the responses and on developing solutions to problems. Otherwise the meeting may degenerate into a gripe session.

It is important to keep in mind that, as with all forms of performance related feedback, that solely negative/critical feedback (e.g., criticizing the employees) is likely to have little positive effect and may have unintended negative consequences. The feedback session should emphasize positive findings as well as negative findings and have development rather than judgment as the primary goal. As noted above, a useful final product for the feedback session is a set of goals for managers and employees that can be periodically assessed and reevaluated if necessary.

Action Plans

Organizational surveys raise employees' expectations concerning the future of the organization. Given this, surveys can foster cynicism if employees perceive that no beneficial changes occur as a result of the survey. The symbolic message sent to employees is, "We have heard you, but we really don't care what you say" If no action plan is intended, then it makes little sense to conduct the survey.

Failure to follow through on a single survey can reduce the effectiveness of a long-term survey research program. Under these circumstances, it takes substantial effort to gain employees' trust enough to conduct an effective survey. Further, the lack of trust fostered by failure to follow through will spread to other elements of the organizational culture.

For surveys to benefit the organization, the following characteristics should be met:

Once a survey has identified areas where change is required, the organization should develop an action plan. Action plans may address relatively minor innovations such as, employee recognition programs, improvements to the company newsletter, or changes to benefits programs. Conversely, action plans may involve sweeping organizational changes.

It is important to note that action plans (and feedback) should reinforce what is done well and emphasize that employees are satisfied and committed to the organization's mission (where this is true).

· The organization must accept data as valid. That is, the directors of the survey project must overcome resistance to the data by various members of the organization. The key to overcoming resistance is to build trust in the project by allowing employees to participate (to the extent that it is feasible) in the development of the survey

· Individual employees must accept responsibility for the part they play in any problems identified in the survey. A properly conducted feedback session should facilitate this process. One key point to keep in mind is that if feedback is solely evaluative and critical, most people are much less likely to accept responsibility for their role in any problems. A simple rule of thumb is that the question, "How can we improve?" is likely to produce a better action plan than the question, "Why did you do that the wrong way?"

· Employees must commit to solving the problems identified in the survey. Individuals' commitment is likely to be affected by the commitment of individuals at higher levels in the organizational hierarchy If senior managers do not stress the importance of the survey to lower level managers, it will be clear to most members of the organization that the results of the survey are not really all that important.

Solving organizational problems usually involves overcoming barriers to change. Many of these barriers can be overcome by careful strategic planning - before conducting the survey. Most effective change strategies involve cultivating commitment from top management, designing the survey to meet strategic organizational needs, building trust from employees before, during, and after data collection by encouraging employee involvement, and maintaining clear lines of communication. The action plan is likely to differ substantially depending on the nature of the change but these principles apply to most change initiatives.

Action plans are more likely to be successful when they involve the articulation of clear, measurable, moderately challenging, and achievable goals that the organization can periodically reevaluate. While it is desirable to set forth general principles for change (e.g., improve the quality of work life), these principles are likely to have little value unless they are tied to specific goals. Articulating the goals for change in the action plan provides employees with a clear sense of the purpose for change and allows them to begin thinking about ways in which the goals may be accomplished.

Finally, expressing measurable goals and then actually measuring them is yet another way of demonstrating commitment to the change effort.

The final phase of the survey project is to clearly communicate the nature of the resulting changes to employees and link them to the survey data.

To maximize the effectiveness of the survey, the organization should show that beneficial changes are occurring and that the changes are a function of employees participation in the survey. As with the feedback process, active employee involvement is the key to effective communication.

The least effective methods involve passive communications such as, posting the changes on the company bulletin board and leaving it to the employees to read the posting and understand the nature of the change. A somewhat active communication method would be personal letters or e-mail messages to employees. Finally, interactive communication methods such team meetings or face to face discussions of the nature of change are likely to the most effective methods. In any case, is critical to link the changes to the survey data.

Surveys can be a one-shot project or a survey research program. Survey programs stress the importance of periodic follow-up on the issues that drive the survey purpose. Whether or not a formal survey research program is devised, it is important to track the evolution of any organizational issue. Therefore, the final step in any survey project is a renewal of search for key topics that might be the focus of survey research or other organizational development efforts.

References Argyris, C. (1993). Knowledge for action: A guide to overcoming barriers to organizational change. San Francisco, CA: Jossey Bass Publishers.

Bass, B.M., Cascio, W.F., & O'Connor, E.J.:Magnitude estimations of expression of frequency and amount. Journal of Applied Psychology, 59, 313-320.

Bernardin, H.J. & Beatty, R.W. (1984). Performance appraisal: Assessing human behavior at work. Boston: Kent.

Berry~ L.L, Parasuraman, A. & Zeithami, V.A.(1994). Improving service quality in America: Lessons learned. Academy of Management Executive, 8, 32-33.

Bracken, D.W. (1996). Multisource (360-Degree) feedback: Surveys for individual and organizational development In A.I. Kraut (Ed.). Organizational surveys: Tools for assessment and change (pp 117-1~). San Francisco: Jossey Bass.

Bradburn, N.M., Sudman, S., Blair, E., & Stocking, C. (1989). Question threat and response bias. In E. Singer, & S. Presser (Eds.). Survey research methods: A reader (pp. 371-384). Chicago: University of Chicago Press.

Burke, W. W., Coruzzi, CA, & Church, XH. (1996). The organizational survey as an intervention for change. In Ad. Kraut (Ed.). Organizational surveys: Tools for assessment and change (pp 41-66). San Francisco: Jossey Bass.

Converse, J.M. & Presser, 5. (1986). Survey questions: Handcrafting the standardized questionnaire. London: sage.

Cummings, T.G. & Huse, Organizational development and change (4th Ed.). St. Paul, MN: West. E.F. (1985).

Diliman, D.A. (1978). Mail and telephone surveys: The total design method. New York: Wiley.

Fox, RJ., Crask, M.R., and Kim, J.H. (1988). Mail survey response rate: A meta-analysis of selected techniques for inducing response. Public Opinion Quarterly, 52, 467-491.

French, W.L. & Bell, C.H., Jr. (1978). Organizational Development Englewood Cliffs, NJ: Prentice Hall.

Higgs, A.C. & Ashworth, S.D. (1996). Organizational surveys: Tools for assessment and research. In A.l. Kraut (Ed.). Organizational surveys: Tools for assessment and change (pp 19-40). San Francisco: Jossey Bass.

Hinrichs, J.R. (1996). Feedback, action planning, and follow-through. In A.I. Kraut (Ed.). Organizational surveys: Tools for assessment and change (pp 255-280). San Francisco: Jossey Bass.

Kuhnert, K. & McCauley, D.P. (1996). Applying alternative survey methods. In A.I. Kraut (Ed.). Organizational surveys: Tools for assessment and change (pp 253-254). San Francisco: Jossey Bass.

Kraut, A.I. (1996). Planning and conducting the survey: Keeping strategic purpose in mind. In A.I. Kraut (Ed.). Organizational surveys: Tools for assessment and change (pp 149-176). San Francisco: Jossey Bass.

Macey, W.H. (1996). Dealing with the data collection, processing, and analysis. In A.I. Kraut (Ed.). Organizational surveys: Tools for assessment and change (pp 204-232). San Francisco: Jossey Bass.

Nadler, D.A. (1996). Setting expectations and reporting results: Conversations with top management In A.I. Kraut (Ed.). Organizational surveys: Tools for assessment and change (pp 177-203). San Francisco: Jossey Bass.

Porras, J.L. & Robertson, P.J. (1992). Organizational development: Theory, practice, and research. In M.D. Dunnette & LM. Rough (Eds.) Handbook of Industrial and Organizational Psychology (2nd Ed., Vol. 3). Palo Alto, CA: Consulting Psychologists Press, Inc.

Schieman, W.A. (1996). Driving change through surveys: Aligning employees customers, and other key stakeholders. In A.I. Kraut (Ed.). Organizational surveys: Tools for assessment and change (pp 88-116). San Francisco: Jossey Bass.

Schieman, WA. (1992). Why change fails. Across the Board. 29, 53-54.

Schmitt, N.W. & Klimoski, R.J. (1991). Research methods in human resource management Cincinnati: South Western Publishing Co.

Schmitt, N.W. & Stults, D.M. (1985). Factors defined by negatively keyed items: The result of car&~s~i respondents? Applied Psychological Measurement, 9, 367-373.

Schuman, H. & Presser, 5. (1996). Questions and answers in attitude surveys: Experiments on question form, wording, and context London: Sage.

Steeh, C.G. (1989). Trends in nonresponse rates, 1952-1979. In E. Singer, & S. Presser (Eds). Survey research methods: A reader (pp. 32-49). Chicago: University of Chicago Press.

Wagner, D.B. & Spencer, J.L. (1996). The role of surveys in transforming culture: Data, knowledge and action. In Al. Kraut (Ed.). Organizational surveys: Tools for assessment and change (pp 67-87). San Francisco: Jossey Bass.

Wiseman, F. (1989). Methodological bias in public opinion surveys. In E. Singer, & S. Presser (Eds.) Survey research methods: A reader (pp. 189-192). Chicago: University of Chicago Press.

Rea, L.M. & Parker, RA. (1992). Designing and conducting survey research. San Francisco: Jossey Bass.

Rodgers, R. & Hunter, I.E. (1991). Impact of management by objectives on organizational productivity. Journal of Applied Psychology, 76, 322-336.