Means of collecting data that does not interfere with the subject's answers - because it is data that already exists, has already been collected by someone else, or does not involve interaction with a subject.
I. Content Analysis
Looking for specific conceptual content in recorded material - including
magazines, records, journals, newspapers, songs, letters handbooks, even
tv or radio broadcasts.
Examples: a) Looking for political content in popular songs
b) looking for racial overtones in children's literature
c) looking at Nursery rhymes - hidden meaning from centuries ago
d) Folk Songs - is a means of public catharsis? Frankie and Johnnie, Banks of the Ohio, Blood on the Saddle
e) Reasons for lynching
f) Sexual activity in 1650s, and 1850s in Europe
g) Incidence of violence in children's cartoons
A. First operationalize the concept into a variable.
1) political content -- reference to political events -- political
symbols -- political sentiment etc.
2) racial overtones -- overt mention of race, use of stereotypes
without mention of race, social status of characteris, etc.
3) public emotion -- guilt, fear, horror, shame
B. What is the Unit of Analysis - the what or whom being studied.
1. UA = political content; UO = popular songs
2. UA = racial overtones; UO = children's literature
3. UA = public emotion, UO = folk songs
Population - the entity to which you want to generalize your study and from which you draw your sample. Has to do with the unit of observation.
Sample - a selected group drawn from your population to represent the population. A probability sample is drawn according to probability theory, typically involving some random selection mechanism.
Types of Samples:
Random sample - selection process by which any member of the population has an equal chance of being selected.
Systematic sample - A type of probability sample in which every kth unit in a list is selected for inclusion in the sample.
Stratified sample - a process of selecting from the population according to specified characteristics so that all groups in the population are represented in the sample.
Quota sample - similar to a stratified sample. Nonprobability sample in which units are selected on the basis of specified characteristics so that the total sample will have the same distribution of characteristics assumed to exist in the population being studied.
Cluster sample - multistage sample in which natural groups (clusters) are sampled initially with the members of each selected group being sub-sampled afterward. I.e. select areas in a city that can be conveniently reached, then randomly select households in those areas to study.
Snowball sample - interview (survey, etc) one person, then request names of other respondents from that interviewee.
D. Decide if your study will be quantitative or qualitative. Most material can be turned into quantitative data, if you are careful. You can use a scale 1= very liberal, 2= moderately liberal, 3= neutral, 4= moderately conservative, 5= very conservative. Be sure that you include a base for the numbers.
Look for disconfirming cases.
If many of them are found, give up your hypothesis, or adjust your hypothesis.
Possibilities of bias can be overcome by: 1) selecting the cases at random; 2) have others review your work; 3) report inconsistencies.
E. Coding = classifying statements or words according to a systematic scheme. Raw data is transformed into standardized form suitable for further analysis.
F. Manifest or Latent content - Are the issues that you are interested in found in the words represented, or are there other issues that you are interested that are not overtly specified. For manifest content you count the number of times that a concept appears. For Latent content you explain why you arrived at your conclusion. However, someone else might arrive at a different conclusion.
G. Is your goal Depth (which has greater validity) or Specificity (greater reliability). The best method is to use both Depth and Specificity (Qualitative and Quantitative).
It is convenient, and easy to do
The data is easy to get
Researcher can redo it easily if there is a major goof in coding, etc.
There is no researcher bias affecting the respondent's response
Disadvantages of Content Analysis
There could be slippage in the recording of the data.
The data may not be exactly appropriate to the subject.
The measures may not be valid.
II.Analyze Existing Statistics
Durkheim's study of suicide.
Martha's study of infanticide
A. Archives - actuarial records, political and judicial records (votes, judgements, etc) city budgets, communications media, government documents (census, etc) International organization records (UN, World Bank)
B. Private records - sales records, industrial and institutional records, written documents, medical records
You don't have individual numbers
Patterns on the group level may not be the same as the individual level
This is called an ecological fallacy
i.e. Durkheim found that more people in Protestant areas commit suicide
than in Catholic areas. But it may be that Catholics in Protestant areas
are the ones committing suicide.
So the conclusions involve inferences -
The existing data may not be exactly the figures that you want.
Logical reasoning can supplement
Replication - the same results are found in a variety of studies
Not all records were kept accurately - laws may change, some things are tabu,
officers, etc are lazy,
Changes in record keeping can alter the incidence of various crimes, etc.
These problems can be countered by knowing all the circumstances.
III.Historical/Comparative Analysis
Historic research enjoyed a resurgence in the late 1970s and 80s. Some have described it less of a particular method than a collection of eclectic techniques and a style of argument. It is research that looks for patterns (cause and effect) in historical material of all types. It is not likely to use large samples, but rather use complex and rich cases.
Examples of Historical research
Religions of China
Religions of India
Edward Shorter - The Making of the American Family
It is important to know how to interpret the historical documents - what biases were common to the period, what was left out, etc.
In 1963 a law was passed requiring public officials to report suspected child abuse. The incidence of reported child abuse increased dramatically.
A. Physical Traces - Accretion - signs of material deposit - dirt, smears,
This thinking can also be applied to other social phenomena,such as the passing of laws, changing of customs, etc.
B. Archival Records: 1) Public Records - actuarial (birth, death, marriage), police and judicial records, government documents, mass media
Also check for fakes. 1) The document doesn't make sense. 2) The document has been copied a lot and there doesn't seem to be an original. 3) There are different versions of the document. 4) Material in the document fits too neatly into accepted theories.
Information in an historical document can be compared to known documents or facts.
Methods of Historical Research
1. Method of Agreement - Find several cases where the outcome is the same, and look for similarities leading up to the outcome.
2. Method of Disagreement - Find cases where the outcome is different, and look at the factors leading up to the outcome.
3. Comparative Historic Method
a) Case Study Approach - has the weakest results because it can't be generalized.
b) Universalizing Method - Look for the common properties in all the instances of the phenomenon.
c) Variation Finding Approach - Look at the differences in a collection of cases. It is hard to find enough data for this approach.
4. Oral Histories - There is a wealth of knowledge to be had, but it is tedious to collect it. Don't take everything you hear at face value. Incidents recalled through memory are always suspect. Look for patterns. Find ways to verify what you are told.