week 7 Flashcards
(57 cards)
1
Q
planning for analysis
A
type of data
type of formatting
type of analysis
2
Q
fomating technique quant
A
- Must “quantify” the data
- Convert (“data reduce”) from collection format into numeric database
3
Q
formatting technique qual data
A
- Must process the data (type/enter/describe)
- Convert from audio/video to text
4
Q
type of data quant
A
- Counts, frequencies, tallies
- Statistical analyses (as appropriate)
5
Q
type of data qual
A
- Coding, categories
- Patterns, themes, theory building
6
Q
whats quantifying data
A
- Before we can do any kind of analysis, we need to quantify our data
- “Quantification” is the process of converting data to a numeric format
- Convert social science data into a “machine readable” form, a form that can be read & manipulated by computer programs
7
Q
examples of quantifying data
A
- Assign numeric representations to nominal or ordinal variables:
Turning male into “1” and female into “2”
Assigning “3” to Very Interested, “2” to Somewhat Interested, “1” to Not Interested - - Assign numeric values to continuous variables:
Turning born in 1973 to “35”
Number of children = “02”
8
Q
developing code categoesi
A
- Some data are more challenging. Open-ended responses must be coded.
- Two basic approaches:
- Begin with a coding scheme derived from the research purpose.
- Generate codes that emerge from the data.
9
Q
goal of coding quant data
A
Goal – reduce a wide variety of information to a more limited set of variable attributes
10
Q
points to remember coding quant
A
- If the data are coded to maintain a good amount of detail, they can always be combined (reduced) later
- However, if you start off with too little detail, you cannot get it back
- If you are using a survey / questionnaire, it’s a good idea to do your coding on the form so that it can be entered properly (e.g., create a “codebook”)
11
Q
purpose of codebook construction
A
- Primary guide used in the coding process.
- Should note the value assigned to each variable attribute (response)
- Guide for locating variables and interpreting codes in the data file during analysis.
- If you’re doing your own input, this will also guide data set construction
12
Q
entering data systems
A
Optical scan sheets (usually ASCII output).
- Limits possible responses
CATI system / On-line: entered while collected
Data entry directly onto an SPSS data matrix, Excel spreadsheet, or ASCII file.
- Typically, work off a coded questionnaire (e.g., data code book based on the questionnaire used)
13
Q
in spss coding
A
- Create variables/column headings & enter each case
- Input data directly into SPSS
- Can also cut and paste Excel file directly into SPSS but have to create variable/column headings in SPSS that correlate with the Excel variable columns
14
Q
whats univariate analysis
A
Frequency distributions
- Measures of central tendency •Mean, Median, Mode
15
Q
subgroup comparisons
A
Describe subsets of cases, subjects or respondents.
Examples
- “Collapsing” response categories: Age categories, Open responses, etc.
- Handling “don’t knows“ Code separately, make missing if appropriate
16
Q
Bivariate analysis
A
Describe a case in terms of two variables simultaneously.
Example:
- Gender
- Attitudes toward equality for men and women
- How does a respondent’s gender affect his or her attitude toward equality for men and women?
Crosstabulations / Correlations
17
Q
multivariate analysis
A
- Analysis of more than two variables simultaneously.
- Can be used to understand the relationship between multiple variables more fully.
- Most typical: Regression analysis
18
Q
data analysis QUANT
A
Descriptive Statistics
- Procedures used to describe a given collection of data.
- The purpose is to describe the sample at hand the collection of cases that we have examined.
Inferential Statistics
- Procedures that let us generalize our findings beyond the particular sample at hand to the larger population represented by that sample.
19
Q
three types of descriptive statistics
A
Central Tendency Measures*
- Variability Measures
- Frequency and Percentages*
20
Q
whats four sleeves of measurement in quant
A
- Nominal: basic classification data; do not have meaningful numbers attached to them, but are broader categories
- Ordinal: have numbers attached to them and the numbers are in a certain order, but there are not equal intervals between the number
- Interval: have equal intervals between the numbers; the distance between attributes have meaning
- Ratio (Scale): have equal intervals between the numbers; there is an absolute zero that is meaningful
21
Q
reporting quant results
A
- Charts
- Graphs
- Tables
22
Q
descriptive statistics examples
A
number frequency count percentage decline and quartiles measures of central tendency (mean, midpoint, mode) variability variance and standard deviation graphs normal curve
23
Q
variability is
and examples
A
differences among scores- shows how subjects vary examples dispersion range variance and standard deviation interval or ration level data
24
Q
whats dispersion
A
extent of scatter around the “average
25
whats variance
spread of scores in a distribution. The greater the scatter, the larger the variance
26
whats standard deviation
how much subjects differ from the mean of their group. The more spread out the subjects are around the mean, the larger the standard deviation. Sensitive to extremes or “outliers
27
inferential statistics
- Allows for comparisons across variables
- i.e. is there a relation between one’s occupation and their reason for using the public library?
- Hypothesis Testing
28
whats level of significance
The level of significance is the predetermined level at which a null hypothesis is not supported. The most common level is p < .05
- P =probability
- < = less than (> = more than)
29
whats type I error
: Reject the null hypothesis when it is really true
30
whats type II error
:Fail to reject the null hypothesis when it is really false
31
P value
- By using inferential statistics to make decisions, we can report the probability that we have made a Type I error (indicated by the p value we report)
- By reporting the p value, we alert readers to the odds that we were incorrect when we decided to reject the null hypothesis
32
chi square test of independence
two variables (nominal and nominal, nominal and ordinal, or ordinal and ordinal)
- Affected by number of cells, number of cases
- 2-tailed distribution= null hypothesis
- 1-tailed distribution= directional hypothesis
33
whats correlation
the extent to which two variables are related across a group of subjects
34
whats Pearson r
- It can range from -1.00 to 1.00
- -1.00 is a perfect inverse relationship—the strongest possible inverse relationship
- 0.00 indicates the complete absence of a relationship
- 1.00 is a perfect positive relationship—the strongest possible direct relationship
- The closer a value is to 0.00, the weaker the relationship
- The closer a value is to -1.00 or +1.00, the stronger it is
- Spearman rho
35
statistical test of difference (t-test)
- Test the difference between two sample means for significance
- pretest to posttest
- Relates to research design
- Perhaps used for information literacy instruction Analysis of variance (ANOVA) tests the difference(s) among two or more means
- Regression analysis (including step-wise regression)
36
graphs for categorical variables
bar diagram and pie chart
37
graphs for numerical variables
Histogram, stem and leaf and Box-plot
38
whats non response
Failure to collect data from a high percentage of those participants selected to be in a sample is a large source of survey error
39
idea in probability theory
is that everyone in a population (or subgroup) will have the opportunity to have data collected about him/herself
40
categories of non respondents
1. Those who the data collection procedures do not reach, thereby not giving them a chance to answer questions.
2. Those asked to provide data who refuse to do so
3. Those who are unable to perform the task or provide the information required of them for some reason (e.g., person does not speak language, some one with poor reading & writing skills, person is too ill to answer questions, person does not have intellectual ability to answer questions, person too young developmentally to answer questions)
41
calculating response rate
- It is simply the number of people responding divided by the number of people sampled.
- usually reported as percentages
42
bias related to non response rate
- The effect of non-response on surveys depends on the % of participants not responding & the extent that those participants not responding are biased.
- Biased non-respondents are systematically different from the whole population
43
whats acceptable response rates range from
25% to 75%.
44
bias associated with non response of mail surveys
- One general rule for mailed surveys is that people who are interested in the subject matter or the research itself are more likely to return mailed questionnaires than those who are less interested.
- This means that mail surveys with low response rates may be biased significantly in ways that are related to the purpose of the survey
- Another consistent bias in mailed surveys is that better educated people often send back mailed questionnaires more quickly than those with less education.
45
bias associated with non response of telephone surveys
- For telephone surveys, availability is an important source of nonresponse.
- If telephone data is collected between 9 am and 5 pm on Sunday Thursday, the people available who are available to be interviewed will be distinctive.
46
groups that tend to be under represented in surveys
- Unemployed
- Single people
- Recent migrants
- People who live in inner city areas
- Low income individuals
- Individuals with low education levels
- People who do not speak English
47
two issues to address to achieve high rate of response
1. Gaining access to the selected individuals.
| 2. Enlisting their co-operation.
48
Reducing Non-response: Due to Lack of Availability
- Make numerous calls during evenings and weekends.
- 6 to 10 calls per household are often needed.
- Have interviewers with flexible schedules who can make appointments at any time that is convenient for the respondents.
49
Reducing Non-response: Enlisting Cooperation
- If possible…send an information letter in advance.
- Effectively and accurately present the purpose of the project.
- Ensure that respondents are not threatened by the task or how the data will be used.
- Have effective interviewers
50
Reducing non response tips:
- It is reasonable to routinely ask people who initially refuse to reconsider participating in a survey.
- A significant % of refusals result from contacting the individual at the wrong time rather than a fundamental unwillingness to be interviewed.
- Approx. .25 to .33 of people who initially refuse to participate to be interviewed will agree when asked again at a later time.
- The survey interview process should be a positive one for respondents.
- Most respondents report that being interviewed is pleasurable…people like the opportunity to talk about themselves to a good listener.
51
Reducing Non-response: Mail Surveys
- Anything makes a mail questionnaire look more professional, more personalised or more attractive will have some positive effects on the response rates.
- Layout of survey should be clear so it is easy to see how to proceed.
- The questions should be attractively spaced, easy to read and uncluttered.
- Response tasks should be easy to do; do not ask respondents to provide written answers, except if they so wish to.
- Response tasks should be to check a box, circle a number or some other equally simple task.
- Some researchers include prepayment of respondents increases mailed response rated.
- The primary difference between good and poor mail surveys is the extent to which researchers make repeated contact with respondents
52
steps to reduce mail non response
- STEPS:
- Mail out surveys.
- About 10 days later mail a reminder postcard.
- About 10 days after postcard mailing then send out a reminder letter to nonrespondents emphasising the importance of the survey topic.
- Last step is to contact respondents via telephone or email.
- In order to track who has responded to the survey and who has not, an identification number can be written on the questionnaire or return envelope.
53
Reducing Non-response: Internet Surveys
- The procedures for internet-based surveys are quite similar to those of mailed surveys.
- Motivated or interested people are most likely to respond.
- Making the task easy, repeating contacts, using more than one mode to contact respondents, and offering alternative modes of responding for those who do not initially respond appear likely to be the keys for maximising response rates.
54
three approaches to minimise the resulting error
1. Using proxy respondents.
2. Doing statistical adjustments.
3. Re-surveying a sample of non-respondents.
55
whats using proxy respondents
- Many surveys routinely collect data from one household respondent about other household members.
- If a respondent is unable or unwilling to be interviewed, asking another household member to report for the designated respondent is one option.
1. Studies indicate that the quality of proxy reporting is not as good as self-reporting for most topics.
2. Is not suitable for reporting information about subjective states such as feelings, knowledge or opinions.
3. Is suitable for factual information.
56
whats doing statistical judgements
- Is possible to weight the answers given by a certain segment of the sample based on the % of the group they represent.
- Statistical adjustments help to reduce the error in estimates caused by non-response.
57
whats resurveying a sample of non respondents
- If a researcher has attained a 60% response rate using a mailed survey, he/she may decide to use another mode of contact and approach a % of the non-respondents using a telephone call to increase the response rate.
