M6 - Survey design Flashcards

1
Q

Survey design problem

A

You cannot talk

–> must be clear; no clarification possible
–> build trust ; ask for sensitive info
–> common-method bias
People tend to tick the answer again

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
2
Q

Survey design elements needed

Where to put them?

A
Welcome - set the tone
Legitimation - importance
Motivation - prices
Story - goal
Trust - sensitive info asked

Where:
invitation letter / survey cover sheet
Instructions on page 1 of the survey
Actual questionnaire

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
3
Q

Steps in a survey project

A
  1. research question & conversation
  2. pop/sample: who will we actually survey on?
    –> the right tone/ customize language
  3. first draft & feedback from
    People not in the survey
    –> length? Sense of questions?
  4. pilot test with people from actual sample
    –> is it clear?
  5. actual survey
How well did you know this?
1
Not at all
2
3
4
5
Perfectly
4
Q

Concept

A

Abstractions communicated through words or other signs that refer to properties among phenomena

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
5
Q

Conceptualization

A

Process to specify what exactly is meant by particular terms

Eg: attitued toward risk, leadership style

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
6
Q

Variables

A

Characteristics of units, taking on diff values, categories, or attributes for diff observations

Eg: age, color

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
7
Q

Types of variables

A
Dichotomous
Discrete
Continuous
Manifest
Latent
How well did you know this?
1
Not at all
2
3
4
5
Perfectly
8
Q

Dichotomous

A

Only 2 possible values

True/ not true

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
9
Q

Discrete

A

A few possible values

AG/ gmbh/ uh

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
10
Q

Continuous

A

Any value within a certain interval

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
11
Q

Manifest

A

Directly observable

Traffic light color

An observable variable (manifest) is an indicator

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
12
Q

Latent

A

Not directly observable

An intent to do sth

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
13
Q

Indicator

What

Should be

A

A rule of correspondance that links a theoretical concept to observable facts

If a variable is directly observable (manifest), the variable is an indicator

Should be:
Unambiguous, understandable, precisely defined
Include all relevant aspects
Appeal to the respondents

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
14
Q

Qualitative scales

A

Nominal (gender, name)

Ordinal (grade in an exam)

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
15
Q

Quantitative scale

A

Interval (temp in degree)
Ratio
Absolute

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
16
Q

Rating scales

A

Likert scale

Guttman scale

Respondent chooses the level that best represents his/her opinion

17
Q

Likert scale

Measures the ….. of ….. to a given ……

Disadv:

A

Measures the degree of agreement on a statement
1-5

Disadv:
Potential bias to the middle

18
Q

Guttman scale

Is a …. of ….. statements
A person who agrees to statement …, should also agree to statement … and ….

A

s a list of escalating statements

A person who agrees to statement 3, should also agree to statement 1 and 2

19
Q

Semantic differential

Measures …. to …. words or statements on a … point scale with …. anchored in ….. adjectives

A

Semantic differential

Measures REACTIONS to STIMULUS measures words or statements on a 7 point scale with ENDPOINTS anchored in OPPOSITE adjectives

20
Q

What is a scale?

Synonym - inventory

A

A number of questions/items that measure a construct

Scale like likert/guttman etc

A scale may habe suv-scales/dimensions

21
Q

Whats KAI?

3 dimensions

A

Kirton-Adaption-Innovation Index

Measures how innovative one is

3dim:
How much individuals like to work with new ideas
How much individuals focus on being efficient
How much individuals confrom tu rules and authority

22
Q

Reliability

What

Types of reliabiliyt

A

How well am i measuring what i want to measure?

Types: 
Test-retest rel.
Parallel test rel.
Split half rel.
Inter-rate rel.
Internal consistency
23
Q

Types of reliability

TR
PT
Sh
IR
IC
A

Test-retest - same test at 2T
Parallel test - corr. Bet. Similar tests at 2T
Split-half: randomly divide sample in half
Inter-rate: for coding; agreement bet. coders
Internal consistency: corr. Bet. Items

24
Q

How to increase validity?

A

Use existing scales
Pilot studies
Good questions
Multiple measurements

25
Q

Where to find existing scales?

A

Research fields
Books
Hint 1 - apply minor tweak
Hint 2: use existing scale

26
Q

Formulating surv y questions 10 rules

A
  1. avoid ambiguous wording
  2. avoid words or expressions the respondent may not know
  3. dont cover more than 1 content with 1 quest. Item
  4. avoid negative instructions
  5. avoid overly short / long sentences
  6. aboid complex questions
  7. avoid abbreviations
  8. aboid suggestive questions
  9. aboid moral obligation/ social desirability
  10. provide the full range of response options
27
Q

Social desirability

A

Tendency to give answers of which the respondant thinks they are socially desirable or approved, rather than accurate and honest

Eg: are you sexist?

28
Q

Acquiescence

A

Tendency to prefer positive over negative answers

Reduce throigh clear and unambigiuous questions

29
Q

Closed questions

Questions phrased …
Answer options are ….

Open end questions are a ………
No ….. desired
Not …… defined

A

Closed questions

Questions phrased PRECISELY
Answer options are DEFINED
–> easier to analyse

Open end questions are a FREE TEXT FIELD
No PRIMING desired
Not PRECISELY defined
–> contain more info

30
Q

Survey tools

Adv
Disadv

What to keep in mind?

A

+ customizable online programs

  • manual entering as source of error
  • data has to be processed to certain format for statistical software

When designing think about coding too!

31
Q

Adv swlf-administered questionnaire

A
Cheaper
Smaller number of skilled analyzers needed
 Fast analysis
Large no. Of participants possible
Anonymity & privacy
32
Q

Disadv self administered questionnaire

A
Stop mid-way
Skip answers
Go through without answering
No clarification possible
Low response rate 
Might represent the most extremes of the pop.
33
Q

Adv researcher administered questionnaire

A

Fewer misunderstanding, incomplete responses

Greater control over the environment

Clarification possible

Higher response rate

34
Q

Disadv researcher adm. quest.

A

Expensive

Large number of skilled staff

Limited numbers

Long time for admin. & analysis