Reference: "SPSS is actually very simple" Author: a foreigner, translation: some Chinese people
, "SPSS statistical analysis based tutorial (Second Edition)" Author: Zhang Wentong
I think: SPSS is a very broken stuff, just the same as Excel. So the only way to keep a diary to learn a little every day, turtle speed forward ......
case processing summary: Case processing summary;
means the program: obtain one or more categorical variables descriptive statistics;
bar chart, the establishment of multi-use classification data.
summaries for groups of cases: Case Summary packet frame;
Histogram: histogram reflects the observed frequency or the number of the continuous variable interval;
scale this part can not read ah. .
2019/12/22
cos overseas salon of 26:
Microsoft analysis and experimental group
A / B test. Simulation
Gangster blog:
http://yixuan.cos.name/cn/
https://statr.me/
strengthen statistical theory
algorithm machine learning;
variable selection; algorithm - the code;
use the code to do the test, implement ideas.
Large data, databases + parallel computing.
Statistical analysis + analysis + Explain the concept;
exp-platform.com
practical significance and significance tests;
distinctions of Xuecheng: Introduction to Computer Science
introduction to computational thinking and datascience
nano-degree data analyst;
a data visualization website:
HTTPS : //flowingdata.com/
2019/12/25
Chapter IV α reliability coefficient
α coefficient value of 0-1, the higher the value, the number of items inside (that is, internal scale) higher consistency.
What scale was? It feels like a questionnaire, but all options are branches of the same item, for example, is: do not agree, in general, agree.
Among them, a question is called a project (variable);
Suppose, is investigating life satisfaction:
In general, ask some (positive items), for example:
I often feel alive? - generally agree disagree;
I always thank others? - do not agree with the general consent;
and so is positive, then ask questions, to understand the first, the reverse will be behind the project.
In the analysis, by sequentially assigned to said branch 1,2,3. Then you can measure the degree of satisfaction of respondents based on a sum of these options. The higher the score, the happier the person, satisfied.
Negative items need to use a reverse coding.
For example: I often feel boring, listless? - do not agree Agree;
it should be used in reverse code: 3, 2,1;
reliability statistics:
Reliability statistics statistical reliability
[alpha] coefficient: The Cronbach [alpha] apos
Item statistics: statistics program, to provide an average value of each item, standard deviation sample size;
scale statistics: scale statistics, provided the total scale mean, standard deviation, sample size;
2019/12/26
Part of inferential statistics, inferential statistics
Overall: Population;
the Sample: sample;
hypothesis testing: hypothesis testing, using a sample of the overall inference;
null hypothesis: null hypothesis;
the alternative hypothesis: alternative hypothesis;
usually, the null hypothesis of no effect is the statement processing ; alternative hypothesis stated process is effective;
unilateral test, One-tailed test;
TWO-tailed test, bilateral test;
sampling errors, sampling error;
P value indicates that the null hypothesis is true, to obtain a particular situation or more extreme results probability;
reject the null hypothesis - referred to as "significantly" or "statistically significant";
not reject the null hypothesis - "not significant";
effect size: effect sizes
2020/1/12
- The value of missing data in the view
only the missing data is numerical value ".";
Missing values in the data string is "" - Menu [help] can obtain the relevant documentation to learn, nothing more than little bit to see;
- Street Intercept: central location test
- In [Variable View missing values in the column may be provided a custom ... missing values
- For special treatment of multiple-choice questions:
multiple choice, also known as "multiple response" (multiple response).- There are two statistical software for multiple choice of storage formats: multiple dichotomy (multiple dichotomy) and multiple classification (multiple category Method);
- Multiple dichotomy: As the name suggests, all options will be set to the variable, if elected, then set the value of 1; if not selected, you set a value of 2. This method can be carried out directly statistical modeling, but a lot of options when the time, efficiency will be reduced;
- Multiple classification: According to the allowable maximum number of options to set the variable, denoted ans1, ans2, ans3, then that can only choose three. Wherein each variable receiving answers respondents, allowed missing values (corresponding to the selected respondents only two). Multiple storage efficiency classification is higher than the multiple dichotomy, however, this method can not be directly carried out statistical modeling, statistical description of the only multiple choice. If you want to statistical modeling, we need to convert multiple classification into multiple dichotomy: [Analysis] - [Multiple response] ...... pending further study;
- Perspective table: Pivot Table.
