Data is data. I was just helping out a graduate nursing student with a study that used the World Values Survey, one of an enormous number of databases with all sorts of data for social scientists (among others). Corpora rely on balanced use of newspapers, novels, TV broadcasts, etc.

That said, not all data can be collected from such sources and in many cases it is difficult to find data in such a way so as to infer much of anything other than that you were able to find what you were looking for. You may find the link attached below helpful. Without knowing the nature of your research question it's hard to say much, but the link goes into the use of news and other similar data for research.

http://gjol.net/wp-content/uploads/2012/12/book-online-research-eng.pdf

Hi Andrew. My study was about about political power and economic stability. My objective is

1. understanding the pattern of economic development and growth that have connection with political decision.

2. study the deficit and economic development:

3. Government loan (internal or external) and relation to the GDP:

4. Study the unstable deficit for the past few years that related to economic stability.

Am I do wrong in my data collection? Could you please suggest how to collect all these data? Am I still need a respondent to collect the data?

Thanks in advance.

Ironic! This is the second question related to complex/dynamical systems approach to economics I've addressed on this site in one day. If I understand you correctly, then newspapers, official reports, etc., are more than adequate (they are ideal). Your issue is less about data than modeling. You have a number of variables that are not only interconnected but extremely complex (you are looking not only at a series of complex interconnected networks but also the evolution of these over time). If you aren't particularly familiar with complex systems, this can still be overcome with a qualitative approach or more modest goals. Were I you, however, I would try to narrow my field of inquiry as much as possible to a single question. One can spend a lifetime "understanding the pattern of economic development and growth" related to politics and political decisions.

I am not an economist, so the only sources I know of offhand that relate your research concern complex systems as applied to economic research. I know there are mountains of datasets available that you can access to obtain information of the type you seek. But most of those I know require access to specialized databases. I can help with questions relating to modeling and complex systems, but I simply do not know enough about economics to point you to specific sources. However, given your research questions I think that data from respondents is actually less useful than the type of data you could collect from public records and media.

Thanks Andrew. That really helpful. I will narrow it down to focus. Because of the complexity of the data that I need to analyse, maybe I will rewrite again my chapter. Thank you once again for your answers. 

Yes. There are study that may be based on documentary evidence. in this case you do not need any respondent.

Hi Andrew, 

Stafs from SPSS is a very good program to analyse open questions.

if you could read portuguese, i have a book that explains how to work with Sftafs and how to analyse qualitative data: Análise categórica, Árvores de Decisão e Análise de Conteúdo em Ciências Sociais e da saúde com o SPSS.

Dear Muhd, I have very few experience with qualitative research. However I guess that main difference in your case is the difference between PRIMARY and SECONDARY data. PRIMARY are the ones which you collect on your own (such as calls, interviews, etc.) - you can control their quality, since you are more familiar with data. On the other hand SECONDARY is anything which you have second-hand, such as data from statistical yearbooks. These data may be very good but often you may not be sure about their quality, reliability...

Dear Dr. Pestana:

I don't have a problem with analyzing qualitative data (far from it). I simply lack sufficient knowledge of economic theory to create the kinds of models necessary here. I must confess also that I really despise SPSS, which (as I've said elsewhere) seems to be designed to perform the maximum number of different kinds of analyses without knowing the underlying logic of any. R, which is vastly more powerful, requires the user to at least write the code, as does SAS, MATLAB, etc. SPSS is mostly Excel with the capacity to easily perform analyses one doesn't understand but get results that one can then claim mean something because (among other things) graduate statistics programs are filled with students who haven't taken linear algebra, yet are now using the GLM (and haven't studied differential equations or other necessary topics in order to understand nonlinear models). SPSS does not have the kinds of tools I prefer to use for qualitative analyses, nor do the research questions involve only qualitative analyses.

For example, consider the stability of any system, economic or no. The standard way to approach modeling this (and most other) systems is via differential equations. Not only can these be extremely complicated and require computational techniques SPSS doesn't offer (numerical integration techniques, for example), but even the basic methods for visually analyzing the phase space of a system compared to CAS systems like Mathematica, still less computational software packages like MATLAB.

I tend to deal with categorical data or qualitative data in general using fuzzy set theory and a number of classification/clustering methods as well as MDS and (less and less) SEM. I have SPSS 22. I can't use most of the methods I generally do with SPSS that I can with R (which is free), or more easily with MATLAB (which has apps for many of these but everything is easily customized because MATLAB is a language not just a computational software platform). I have STATA, Minitab, Mathematica, Maple, R, & SAGE in addition to MATLAB and SPSS. I have found that the only time I use SPSS is when I have Excel files that I would have to read in to e.g., MATLAB but can readily load into SPSS and my only desire is simple graphical displays.

The issue when it comes to analyzing popular media, finding particular databases, etc., generally involves methodological issues that come before analysis (a big exception being data mining and so forth, which SPSS is of little use for but MATLAB, R, and other software packages excellent for). It is true that coding variables the way that is often necessary in qualitative analyses that involve datasets obtained from newspapers or similar sources is generally simpler in SPSS. However, having coded such data, one is then limited by the limitations of SPSS (which are extensive). That said, it isn't that hard to output a coded dataset from SPSS such that I can then use it for more sophisticated model construction or analyses using MATLAB.

All of your objectives(except objective no 1) clearly contain answer to your query.There is no need of respondents as it is not a survey .In-depth interviews from experts and officials,  however, seem to be appropriate tool of data collection for your study, beside documents you have mentioned. Good luck

Hi Asgher, thank you. Do you have any idea how to analyze the data without respondent? which method in qualitative is appropriate to my research?

I have never analyse a data without interview. Usually I use a primary data since I did not get a respondent I need to analyze a secondary data. Which one above is reliable to my research? 

Thank you for your kind help.

Many of your research questions concern models of economic growth or the evolution of a particular set of variables in an economic model. They are quite quantitative. For example, the relationship between government load and GDP involves (fairly) straightforward numerical data. These can be analyzed using SEM, standard statistical methods from the general linear model (GLM), and sophisticated methods that borrow a lot from machine learning. The stability of any system, whether it concerns grades for undergraduate in a specific course, the climate, or ideal gas laws and thermodynamics, is more or less the same: it requires a dynamical systems approach (a model of the system with the relevant variables such that the model can show how the system evolves over time and the extent to which it is stable or unstable).

Your topics are extremely broad. I would suggest starting smaller (a single question that you have some idea as to the relevant variables that would encapsulate or define the phenomenon/system in question). After narrowing your topic, you can then go back to the ideal way to garner data and to analyzer it. I can direct you to open source media (textbooks, tutorials, seminars, even entire courses) if you wish that cover all these issues and more and from which you can select material that seems relevant to your project. For example, the Open Textbook Library and iTunesU contain multiple sources of various kinds that are relevant to your inquiries.

HI Andrew. Any suggestion to narrow down my research objective? 

Read more on Systematic Review of COntent Analysis

Hi  Afiq,

Normally these sources of data are not reliable as these reports and official data are highly contextual and one can not generalize the results based on these  data. you may get general results but  it depends what are goals and end objectives. 

regards

Rafique

To answer your basic question, yes, you can use reports or any written documents as a source for a research study.  The technique for analyzing this type of data is called content analysis.  Here's a good website that describes the process.  It also includes information about using software.  http://www.contentanalysis.org/  There are many books written on the process.

As has been discussed the validity of the sources of reports etc needs to be considered and described in your research study.  If you are going to do a meta analysis of the data, combining quantitative data from the different sources, then you do have a quantitative study.  If you are analyzing the information in paper sources for trends or more qualitative type questions, this can be a qualitative study.  In the US, for example, women still make less money then men doing the same work.  This varies from state to state.  Reactions to this trend, the fields of study in which this happens the most,  explanations for this trend are qualitative in nature.   It depends on your research question. 

The responses already listed above are good. You can you public open access data - social media and other sources of media... I depends what your research question is of course (as others have mentioned) - you could further refine you content analysis to explore positive  or negative sentiment perhaps... so long as you have a clear chain of logical through out your research, you have a great deal of flexibility in regard to how  you answer you question. Planning you research design is worth spending some time on... and then to seek peer review. Also - be mindful of how you will address any ethical issues. You need to plan this as well - even with public domain data. Best wishes!

I would suggest you look closely into the qualitative literature on content analysis as that is precisely what this method will enable you to do.Delving into text, however it is compiled and whatever the source may be, can be analyze to yield substantive findings. Analysis of this data with this method follows the same rigor as other forms of qualitative analysis so stay true to the method and school of thought. 

I used to evaluate newspaper articles for social research purposes, for example a day  to day journal of a local election......content analsys. Focusing on one media source, one topic, one author, one research question. (Just an example) - The last evaluation I conducted took the form of a daily checklist which I tabulated and analyzed at the end of the election. I was searching for author biases toward particular local issues (You need to research validity issues or find valid templates that others have used)

Evaluation tips for articles:

http://bcs.bedfordstmartins.com/resdoc5e/tips-for-evaluating-sources.htm

View attached article, lots of templates:

The responses already listed above are good.

Content analysis is the most appropriate, but you must specify the criteria for the research sample design for greater validation. For example, if you analyze secondary data as interviews in media, you must specify the: time period  (last week/month/quarter/year etc); type of media; units of analysis, coding...etc. 

A categoria desporto resulta do patente por incluir as várias modalidades nele especificadas; enquanto a categoria estilo de aprendizagem profundo é latente por incluir uma interligação de conceitos (Newble e Clarke, 1986).

Com base nos recursos linguísticos, e após vários refinamentos, criam-se as categorias que capturem todas as ideias contidas nas respostas, esgotando a totalidade do texto.

Os recursos linguísticos englobam dicionários, que constam de várias bibliotecas pré--existentes ou construídos pelo analista e devem ser aperfeiçoados de forma a tornar as categorias mais fiáveis e precisas.

Os refinamentos ou melhorias a introduzir, consistem na definição de sinónimos, de modo a agrupar termos semelhantes, nomeadamente prefixos e sufixos, maiúsculas e minúsculas, singular e plural; na definição de vocábulos irrelevantes a excluir, como verbos e preposições; na criação de tipos que agreguem termos semelhantes; no forçar palavras ou frases não extraídas a fazerem parte de uma categoria; no criar regras condicionais de associação ou de substituição.

Aquando da extracção dos resultados, os termos aparecem seguidos das respectivas frequências, apresentadas entre parêntesis. Normalmente, à maior frequência corresponde a maior importância atribuída pelos respondentes mas, por vezes, há excepções que devem ser objecto de reclassificação.

As categorias representam o nível mais elevado de conceitos que captam as ideias chave ou as atitudes expressas pelos respondentes e podem conter termos, tipos ou padrões.

Os termos (Term) são as unidades básicas de classificação, como por exemplo: leitura, piano, escutismo.

Os tipos (Type) são agregados de ordem superior contendo termos que traduzem a mesma ideia. São exemplos, o tipo desporto que engloba diferentes modalidades cujos termos podem ser o ténis, a natação, o xadrez; o tipo positivo que engloba adjectivações favoráveis caso de frio em relação aos gelados; o tipo negativo que engloba adjectivações desfavoráveis como o frio referido a tomar banho; o tipo orçamento, que engloba os termos custos, preços, comissões, taxas.

Os tipos desconhecidos referem-se a termos omissos nos recursos linguísticos.

O Text Analysis engloba tipos (Types) pré-definidos que não podem ser alterados tanto na Core Library como na Opinions Library.

A Core Library inclui cinco tipos: pessoa, produto, localização, organização e desconhecido.

A Opinions Library abrange 4 tipos de qualificadores (contextuais, positivos, negativos, incertos) e três tipos designados por negativos, positivos ou incertos. Os padrões são combinações de termos ou de tipos e representam comentários positivos, negativos, ou neutros, sobre um determinado termo ou tipo. Para criar categorias nem sempre é necessário recorrer a níveis superiores de contextualização, caso dos termos guitarra, piano e concertos, os quais criam directamente a categoria música.

Havendo ambiguidade na identificação dos elementos codificados, devem definir-se unidades de contexto superiores que estabeleçam critérios susceptíveis de fazer emergir um sentido nos itens obtidos, organizando-os no seu contexto.

A análise de conteúdo faz-se em três momentos:

(1) No primeiro momento acede-se à base de dados que pode ter origem num dos seguintes quatro ficheiros: SPSS, Excel, ODBC e Dimensions.

A origem do ficheiro em nada altera o tratamento estatístico dos dados, pelo que aqui se optou por importar um ficheiro do SPSS, cuja extensão é .sav.

Este ficheiro para além das respostas abertas e de outras variáveis consideradas relevantes para a sua interpretação, deve incluir, obrigatoriamente, uma variável que identifica o número do questionário.

A variável ID é indispensável não só para aceder ao Text Analysis que cria as categorias associadas às respostas abertas, mas também para exportar essas categorias para o SPSS.

O ID pode criar-se de forma expedita no SPSS accionando o comando Tansform Compute como se mostra:

Uma vez que o Text Analysis não permite emendas nas respostas abertas, obriga ao seu tratamento prévio no ficheiro inicial .sav, o qual consiste na:

(i) Correcção dos erros ortográficos.

(ii) Junção numa só variável de diferentes respostas múltiplas que versam o mesmo assunto, admitindo que as variáveis são nominais. Por exemplo, juntar os conteúdos das três principais opções de ocupação dos tempos livres, que constam de três colunas, OTL1, OTL2, OTL3, numa só OTL.

Caso interessasse distinguir a ordem com que são enumeradas essas opções, não faria então sentido o seu agrupamento.

(iii) Substituição pelo termo não resposta aquando da abstenção dos respondentes.

As abstenções, após serem identificadas com uma pelica, são substituídas no SPSS rapidamente pela expressão não resposta, seleccionando-se a resposta aberta. 

No segundo momento da análise, acede-se ao Text Analysis, cujos ficheiros têm a extensão .tas, importando o ficheiro do SPSS com os dados já emendados.

A finalidade desta etapa consiste na criação de categorias associadas às respostas abertas, as quais podem ser obtidas recorrendo iterativamente ao ecrã Text Analysis view (ponto 8.2) e ao écrã Dictionary Editor view (ponto 8.3).

Após a criação das categorias, o ficheiro tas é exportado e salvo num ficheiro do SPSS denominado texto sav, que apenas contém o ID e as categorias criadas.

(3) No terceiro momento regressa-se ao ficheiro inicial do SPSS, que contem as respostas abertas não tratadas e adiciona-se-lhe as categorias guardadas em texto.sav, recorrendo ao comando Merge Variables.

Your search will be limited to "literature" with some statistical issues from data sources that you willing. In this case, you must necessarily address the limitations as to document data sources as device of data bases in statistical methodology, since all the information you would like to have are not fully included.

Hello Muhd

I understand that you are wanting to undertake qualitative data collection and analysis. I’m unsure as to whether you are doing this as part of a mixed-method study, but I can offer some ideas.

Certainly, you can use a range of organisational and other texts as data. Using organisational texts is a key method used in institutional ethnography. The work by Dorothy Smith may be useful to you here.

You may find these references useful, although she is widely cited elsewhere:

• Smith, D. (2001); Text and the Ontology of Organisations and Institutions. Studies in Cultures, Organisations and Societies. Vol. 7, pp 159-198
• Smith, D. (2005) Institutional Ethnography: A Sociology for People. Toronto: AltaMira Press

You mention that you are also interested in ‘power’. Studying institutional texts can offer valuable insights into how power circulates in organisations, and what power ‘does’ when it is exercised.

But, yes, using texts of various kinds – documents, newspaper articles, websites, etc. – are sources of data in some kinds of qualitative research; it just depends on what question you are trying to answer, and your theoretical position.

yes u continue your research on secondary data  

Thank you all of you. That really open my mind to explore others possibility. Am I still could do a Mix method in a really short of time (until June 2015)?

Actually, without respondents, it's all up to you, more control, so yes, absolutely.

There are a range of software packages that can also help. Some are free to use. Here is a web article that might help: http://www.predictiveanalyticstoday.com/top-11-free-software-for-text-analysistext-mining-text-analytics/

Good luck!

possible?   -- yes

reliable? -- 'maybe or not really.. you should provide the research question and objective.. Either you modify your RQ and RO to align with the secondary literature data that you have.... or expand your secondary literature to suit the RQ & RO.

Going back to the original research, it might be that your research design is not right or needs to get a different set of respondents.

The analysis of written documents is not only reliable, but is an accepted method of qualitative research. Reliable is dependent on the how and what of the use of this method.  In quantitative research, secondary data analysis is reputable, reliable and used quite often. As the other, reliable depends on theda ta used and how it is used analyzed..

"... official report, statistics, newspaper article or interview by a newspaper?" Any of these alone or in combination are acceptable and reliable.  HOWEVER, if you are unfamiliar 1) how it can be done, (let alone) that 2) it can be done, perhaps you need to consult with your advisor or review , or, take a course on qualitative methods. 

Many new to research believe that interviewing IS qualitative methods and is the only method to use, thus, they design such a project w/o full knowledge of qualitative research and its  methods.

Hi Mary,

Most of the research (thesis) in my university is mostly by interview for qualitative design. Even the class I attended urged all the student to use a interview as part of the research an add some of secondary data. After all the answer given, now I am confident to use a secondary data that I gather from article for a two years. 

Hi Harry,

That truly right. I need do a major change in chapter 1 to align with a secondary data.

Thanks all of you. It really help me a lot. 

Regards,

Afiq