Survey Design
Level 1 Evaluations
Kirkpatrick Level 1 evaluations, commonly known as "smile sheets," play a crucial role in assessing participants' reactions to their learning experiences. When constructing an effective evaluation, it's essential to diversify the questions asked. This diversity should encompass both question types and categories.
Incorporating a range of question formats, such as Likert scale queries, open-ended prompts, and multiple-choice items, serves several purposes:
- Mitigating Response Biases: by employing different question styles, we minimize the risk of response biases, thereby obtaining more impartial and accurate participant feedback.
- Offering a Holistic Perspective: diverse question types provide a more comprehensive understanding of participant reactions, enabling us to capture both positive and negative feedback.
- Spotlighting Areas for Enhancement: these varied questions help identify specific areas for improvement within the learning experience.
- Enhancing Participant Engagement: by diversifying the survey format, we keep participants engaged and reduce survey fatigue, resulting in more meaningful responses.
Below are sample questions you might consider incorporating into your evaluation:
Category |
Sample Questions |
---|---|
Overall Satisfaction |
|
Intent to Apply (i.e., scrap learning) |
|
Net Promoter Score |
|
Impact |
|
Content Relevance |
|
Delivery |
|
General Comments and Suggestions |
|
General Guidelines: Writing Survey Questions
When writing survey questions, consider the following:
- Because words influence how people perceive things, create a numeric scale and anchor only the ends with words. For example, a Likert scale of 1-7 should only have “1 = strongly disagree” and “7 = strongly agree.” This allows the respondent to interpret the scale as opposed to the writer providing definitions.
- Use odd-numbered scales. 7-, 9-, and 11-point scales are best because they provide the respondent with more choices and allow them to pick a neutral response (4, 5, and 6) where necessary. Use an even-numbered scale when you want to force respondents “off the fence.”
- Keep question sections (i.e., sets) unlabeled; labels can introduce bias (depending on the survey).
- Word about 1/3 of survey questions so that the desired answer is negative. This can create more valid data responses because it forces the respondent to pay attention to each question instead of just marking everything as “strongly agree.”
- As a best practice, write all items positively and then find the easiest ones to turn negative. For example, turn “My manager gets to know people” into “My manager doesn’t get to know people.”
- Focus on observable behavior; if you can’t see it, don’t ask. For example, the Likert statement “My manager considers whether or not they are knowledgeable about their job(s)” does not account for observable behavior (i.e., “knowledgeable about their job(s)” is not observable).
- Design sections to contain a similar number of items and questions that contain a similar number of words; keep comparable sizes close together throughout the survey.
- Limit items to a single description of behavior. For example, the Likert statement “My manager provides good feedback and avoids getting emotional” contains two behaviors. Depending on how the reader interprets the questions, it could introduce bias.
- Keep demographic information to the end.
- Remove the highest and lowest scores from each question/data set when tabulating results.